Differences In Case Mix Research Articles

Abstract WP158: Validation of a Preliminary Prediction Model with MR Plaque Imaging to Estimate Risk for New Ischemic Brain Lesions after Carotid Endarterectomy or Stenting

[Objective] Recently, we reported a preliminary prediction model with carotid plaque MRI to estimate risk for new ischaemic brain lesions after CEA or CAS. The objective of this study was to validate this model in new set of patients with carotid stenosis. [Methods] One hundred four patients with carotid stenosis undergoing treatment (63 CEA, 41 CAS) were used as a training set for construction of a preliminary prediction model to estimate risk for new ischemic brain lesions after CEA or CAS. T1 and T2 signal intensity of carotid plaque were measured on black-blood MRI. Associations among MRI findings, treatment, clinical factors, and occurrence of new ischemic lesions on DWI 1 day after treatment were studied by logistic regression. The validity of the prediction model was examined using a new set of patients with carotid stenosis (n = 43) as a validation set. [Results] In the training set, new DWI lesions after treatment were observed in 25 patients (24%). The model demonstrated that T1-signal intensity and CAS were positively associated with new lesions on post-treatment DWI scans, and T2 signal intensity was negatively associated (Fig. 1). The C-index was 0.79, which indicated some predictive value. In the validation set, new DWI lesions after treatment were observed in 10 patients (23%). However, C-index was 0.6 and positive predictive value was 33% (Fig. 2), which suggested overfitting of our model and/or differences in case-mix between the training set and validation set. [Conclusions] Our preliminary prediction model may provide some useful information for decision-making regarding treatment strategy, but needs further collection of patients to improve its predictive value.

Use of surgical-site infection rates to rank hospital performance across several types of surgery

Comparing and ranking hospitals based on health outcomes is becoming increasingly popular, although case-mix differences between hospitals and random variation are known to distort interpretation. The aim of this study was to explore whether surgical-site infection (SSI) rates are suitable for comparing hospitals, taking into account case-mix differences and random variation. Data from the national surveillance network in the Netherlands, on the eight most frequently registered types of surgery for the year 2009, were used to calculate SSI rates. The variation in SSI rate between hospitals was estimated with multivariable fixed- and random-effects logistic regression models to account for random variation and case mix. 'Rankability' (as the reliability of ranking) of the SSI rates was calculated by relating within-hospital variation to between-hospital variation. Thirty-four hospitals reported on 13 629 patients, with overall SSI rates per surgical procedure varying between 0 and 15·1 per cent. Statistically significant differences in SSI rate between hospitals were found for colonic resection, caesarean section and for all operations combined. Rankability was 80 per cent for colonic resection but 0 per cent for caesarean section. Rankability was 8 per cent in all operations combined, as the differences in SSI rates were explained mainly by case mix. When comparing SSI rates in all operations, differences between hospitals were explained by case mix. For individual types of surgery, case mix varied less between hospitals, and differences were explained largely by random variation. Although SSI rates may be used for monitoring quality improvement within hospitals, they should not be used for ranking hospitals.

Defining ventilator-associated pneumonia: a (de) construction concept

Such variability comes mainly from two aspects: the presence of case-mix differences among the different units evaluated in the literature and the lack of accurate diagnostic criteria that allow for an accurate functional diagnosis, making subjectivity an important aspect of definitive diagnosis and treatment decisions.

Efficacy of biological agents in juvenile idiopathic arthritis: a systematic review using indirect comparisons

ObjectiveOver the past decade, the availability of biological agents for the treatment of juvenile idiopathic arthritis (JIA) has increased substantially. Because direct head-to-head trials comparing these agents are lacking, we...

Abstract 18746: Variation in Intravenous Inotrope Use for Patients Hospitalized with Heart Failure: Data from Get withthe Guidelines

Background: Data guiding the use of intravenous inotrope agents for patients hospitalized with heart failure (HF) are sparse. Guidelines crudely recommend inotrope use for highly selected patients with decompensated HF, hypotension, and evidence of end-organ hypoperfusion. Real-world patterns of inotrope use are not well characterized. Methods: We examined inotrope use for patients hospitalized with HF from 2005-2011 and enrolled in the Get With The Guidelines (GWTG-HF) registry. We used hierarchical logistic regression models to estimate risk-standardized hospital-level utilization rates of dobutamine, dopamine and milrinone, adjusting for clinical and demographic factors. Results: Overall, 7,909 of 130,224 (6.1%) of HF patients hospitalized at 240 GWTG participating centers received inotrope administration (dobutamine 44%, dopamine 24%, milrinone 17%, combination 15%). Rates of inotrope use were constant from 2005 to 2011. At the hospital level, risk-standardized institutional rates of inotrope rates use ranged from 0-34% (median 5.1%, 25th-75th percentile 2.0-9.9%, Figure). Hospital region, rural location, size, and academic status were not significantly associated with the use of an inotropic agent. Conclusion: Inotrope use during hospitalization varied markedly across participating hospitals. This degree of institutional variation, even after accounting for measured differences in case mix, suggests use patterns that are based to a greater degree on individual site propensity than by patient factors. The implications of these markedly different patterns of care are worthy of further investigation .

Trends in consultant clinical activity and the effect of the 2003 contract change: retrospective analysis of secondary data

To explore trends in the clinical activity of hospital specialists in English National Health Service (NHS) hospitals, and test the effect of contract reform. Retrospective secondary analysis of hospital episode statistics, describing trends and testing for a contract effect using multilevel interrupted time series analysis. Ten surgical and medical specialties in English NHS hospitals from 1999 to 2009. Participants Hospital consultants on full time or maximum part-time English NHS hospitals. A new contract offered to NHS consultants in October 2003, with higher pay alongside job planning and appraisal. Inpatient finished consultant episodes (FCEs) per consultant per month, with and without accounting for case-mix differences. In most specialties there was a statistically significant downward trend in FCEs per consultant per month. On average in the surgical specialties, FCEs reduced by 0.14 per month (95% CI -0.16 to -0.11) and in medicine there was a smaller reduction of 0.08 FCEs per month (95% CI -0.1 to -0.06). NHS patients symptoms have increased in severity over time, and the downward trend is reduced after case-mix-adjustment, and reversed in general and geriatric medicine. The effect of the contract on clinical activity was minimal. In five specialties there was no statistically significant effect, but in five specialties there was a statistically significant negative effect. Consultant clinical activity, as measured by FCEs per month, has shown a general downward trend from 1999 to 2009. The consultant contract was not associated with an increase in consultant clinical activity rates.

PSU19 Does the Novel Renal Marker NGAL (Neutrophil Gelatinase-Associated Lipocalin) Result in Cost Savings in Patients With Renal Impairment Following Cardiac Surgery?

Cardiac surgery can result in renal impairment necessitating hemofiltration (HF; dialysis) which prolongs the length of therapy (LOT), days in the intensive care unit (ICU) and is associated with an overall higher mortality. The decision to initiate HF-therapy is currently made on increased plasma creatinine levels, a relatively late marker (peaking days after a potentially reversible renal dysfunction). The novel urine marker NGAL (neutrophil-gelatinase-associated lipocalin) measures the acute structural damage and identifies the need for HF within hours with high accuracy. Based on data collected in 2005-2009, we prospectively included NGAL in an algorithm to modify the management of HF-therapy in 2010. Changes in incidence of HF and LOT, all costs associated with routine NGAL-testing and ICU stay (direct and indirect) as well as the economic impact were analysed. A total of 528 patients were operated with no statistical difference in numbers, case mix, risk scores and outcomes to the previous years. The expected number of patients with HF was 45.3 with an average LOT of 7.7 days (2005-2009). Introduction of NGAL-testing added costs of € 60.00/patient (€ 31.680/year). Although € 2047,50/day could be realized through public health refunding, the fixed associated costs were € 3249,70/patient/day [€ 892,52 for all supplies to operate the HF (Prismaflex, Gambro, Germany) and € 2357,18 for institutional ICU expenditure (incorporating personnel costs)] resulting in a deficit of € 1202,19/day with, and € 751,58 without HF operating. Using the new marker, only 32 patients required treatment with HF for a mean of 5.5 days and earlier initiation (1.1 days) in 2010 (p<0,001). A total of 41.6 days of treatment in ICU (occupancy) were avoided with savings of € 44.153,16. Routine NGAL testing and consecutive changes in management reduced the number of HF-therapies, lengths of therapy and ICU stay in our institution and proved to be cost-saving.

Comparing measures of multimorbidity to predict outcomes in primary care: a cross sectional study

Background.An increasing proportion of people are living with multiple health conditions, or ‘multimorbidity’. Measures of multimorbidity are useful in studies of interventions in primary care to take account of confounding due to differences in case-mix.Objectives.Assess the predictive validity of commonly used measures of multimorbidity in relation to a health outcome (mortality) and a measure of health service utilization (consultation rate).Methods.We included 95372 patients registered on 1 April 2005 at 174 English general practices included in the General Practice Research Database. Using regression models we compared the explanatory power of six measures of multimorbidity: count of chronic diseases from the Quality and Outcomes Framework (QOF); Charlson index; count of prescribed drugs; three measures from the John Hopkins ACG software [Expanded Diagnosis Clusters count (EDCs), Adjusted Clinical Groups (ACGs), Resource Utilisation Bands (RUBs)].Results.A model containing demographics and GP practice alone explained 22% of the uncertainty in consultation rates. The number of prescribed drugs, ACG category, EDC count, RUB category, QOF disease count, or Charlson index increased this to 42%, 37%, 36%, 35%, 30%, and 26%, respectively. Measures of multimorbidity made little difference to the fit of a model predicting 3-year mortality. Nonetheless, Charlson index score was the best performing measure, followed by the number of prescribed drugs.Conclusion.The number of prescribed drugs is the most powerful measure for predicting future consultations and the second most powerful measure for predicting mortality. It may have potential as a simple proxy measure of multimorbidity in primary care.

110 Risk Stratification in Thoracic Aortic Disease: Does Aortic Size Correspond to Medial Degeneration?

more likely to be older and female, to have a BMI 25kg/m, to have more co-morbid illness, to have had prior cardiac surgery and to present with greater surgical acuity. Intra-operatively, they were more likely to have had a lower pre-bypass hematocrit, to have undergone more complex procedures, and to have experienced prolonged bypass times. When practice pattern characteristics were considered, significant variability in peri-operative transfusion rates was noted between surgeon (from 32.4% to 51.5%, p 0.0001) and anesthesiologist (from 34.4% to 51.9%, p 0.0001) and across year (28.2% in 2004 to 48.8% in 2008, p 0.0001). Following adjustment for differences in clinical characteristics and case mix, surgeon, anesthesiologist and year all emerged as independent predictors of peri-operative transfusion (Table 1). CONCLUSION: This study highlighted the potential impact that practice pattern characteristics had on variability in peri-operative transfusion rates within a single center. When attempting to identify mechanisms by which to reduce rates of transfusion rates in cardiac surgery, attention must be paid to transfusion practices that may differ between various practitioners of a cardiac surgical team.

The impact of obstetric unit closures on maternal and infant pregnancy outcomes.

To define the association between large-scale obstetric unit closures and relative changes in maternal and neonatal outcomes. Birth and death certificates were linked to maternal and neonatal hospital discharge records for all births between January 1, 1995 and June 30, 2005 in Philadelphia, which experienced the closure of 9 of 19 obstetric units between 1997 and 2005, and five surrounding counties and eight urban counties that did not experience a similar reduction in obstetric units. A before-and-after study design with an untreated control group compared changes in perinatal outcomes in Philadelphia to five surrounding control counties and eight urban control counties after controlling for case mix differences and secular trends (N = 3,140,782). Relative to the preclosure years, the difference in neonatal mortality (odds ratio (OR) 1.49, 95 percent CI 1.12-2.00) and all perinatal mortality (OR 1.53, 95 percent CI 1.14-2.04) increased for Philadelphia residents compared with both control groups between 1997 and 1999. After 2000, there was no statistically significant change in any outcome in Philadelphia county compared with the preclosure epoch. Obstetric unit closures were initially associated with adverse changes in perinatal outcomes, but these outcomes ameliorated over time.

The importance of audit to monitor applications of procedures and improve primary angioplasty results

Although clinical trials have demonstrated that primary percutaneous coronary intervention (PPCI) provides better outcomes than thrombolysis for STEMI, it cannot be assumed that similar results can be obtained in day-to-day practice. To determine whether standards are being met, continuous audit of PPCI programmes is necessary, with appropriate feedback to participating centres and operators. Both the MINAP and BCIS national audit projects allow central electronic collection of data on consecutive patients presenting to every hospital involved in the acute management of these patients. Regular programmed feedback is provided to centres performing primary PCI that attempts to take account of statistical variation and differences in case mix between units by making use of funnel plots, statistical process control graphs and risk adjustment models. This reporting of "process" and "outcome" data, both confidentially and within the public domain, has been used to drive up clinical performance and has been associated with steady improvements and reduced inequalities of care.

Efficacy and timing of intra-aortic counterpulsation in patients with ST-elevation myocardial infarction complicated by cardiogenic shock

Guidelines strongly recommend additional intra-aortic balloon pump (IABP) therapy in STEMI patients with cardiogenic shock (CS) treated by primary percutaneous coronary intervention (PCI). However, there is no randomised evidence suggesting survival benefit of IABP treatment in CS. It is suggested that timing of initiation of IABP therapy could be of great importance. Therefore, we compared mortality rates of IABP therapy versus no IABP therapy in the setting of STEMI complicated by CS. In addition, we investigated the effect of initiation of IABP therapy on mortality. From a cohort of 292 STEMI patients with CS treated by primary PCI, 199 patients received IABP therapy (IABP group) and 93 patients received no support (no IABP group). The IABP group was divided into two subgroups based on timing of initiation of support, i.e. 'IABP pre PCI' (n = 59) and 'IABP post PCI' (n = 140). Outcomes were assessed by propensity stratification and multivariate logistic regression. All-cause 30-day mortality for the IABP versus the no IABP group was 47% vs. 28%, respectively, in univariate analysis resulting in an odds ratio (OR) of 1.67 (95%CI, 1.16 to 2.39). However, analyses adjusting outcomes by propensity stratification and logistic regression, respectively, neutralised this OR. In the IABP pre-PCI group vs. the post-PCI group 30-day mortality was 64% vs. 40%, resulting in an OR of 1.56 (95% CI, 1.18 to 2.08). However, after propensity stratification analysis and multivariate logistic regression analysis, there were no significant differences in odds of 30-day mortality. In our cohort of patients with STEMI complicated by CS treated with primary PCI we observed a difference in mortality between those treated with IABP and those treated without IABP in favour of the 'no IABP' group. The mortality difference was eliminated after adjustment for differences in case mix by propensity stratification or by logistic regression analysis. Neither did we observe any difference in mortality between patients whose IABP treatment was initiated before or immediately after PCI.

Variation in Intracranial Pressure Monitoring and Outcomes in Pediatric Traumatic Brain Injury

To describe between-hospital and patient-level variation in intracranial pressure (ICP) monitoring and to evaluate ICP monitoring in association with hospital features and outcome in children with traumatic brain injury (TBI). Retrospective cohort study. SETTING Children's hospitals participating in the Pediatric Health Information System database (January 2001 to June 2011). Children (aged <18 years) with TBI and head Abbreviated Injury Scale scores of at least 3 who were ventilated for at least 96 consecutive hours or who died in the first 4 days after hospital admission. Monitoring of ICP. A total of 4667 children met the study criteria. Hospital mortality was 41% (n = 1919). Overall, 55% of patients (n = 2586) received ICP monitoring. Expected hospital ICP monitoring rates after adjustment for patient age, cardiac arrest, inflicted injury, craniotomy or craniectomy, head Abbreviated Injury Scale score, and Injury Severity Score were 47% to 60%. Observed hospital ICP monitoring rates were 14% to 83%. Hospitals with more observed ICP monitoring, relative to expected, and hospitals with higher patient volumes had lower rates of mortality or severe disability. After adjustment for between-hospital variation and patient severity of injury, ICP monitoring was independently associated with age 1 year and older (odds ratio, 3.1; 95% CI, 2.5-3.8) vs age younger than 1 year. There was significant between-hospital variation in ICP monitoring that cannot be attributed solely to differences in case mix. Hospitals that monitor ICP more frequently and hospitals with higher patient volumes had better patient outcomes. Infants with TBI are less likely to receive ICP monitoring than are older children.

Injury profiles related to mortality in patients with a low Injury Severity Score

Outcome prediction models are widely used to evaluate trauma care. External benchmarking provides individual institutions with a tool to compare survival with a reference dataset. However, these models do have limitations. In this study, the hypothesis was tested whether specific injuries are associated with increased mortality and whether differences in case-mix of these injuries influence outcome comparison. A retrospective study was conducted in a Dutch trauma region. Injury profiles, based on injuries most frequently endured by unexpected death, were determined. The association between these injury profiles and mortality was studied in patients with a low Injury Severity Score by logistic regression. The standardized survival of our population (Ws statistic) was compared with North-American and British reference databases, with and without patients suffering from previously defined injury profiles. In total, 14,811 patients were included. Hip fractures, minor pelvic fractures, femur fractures, and minor thoracic injuries were significantly associated with mortality corrected for age, sex, and physiologic derangement in patients with a low injury severity. Odds ratios ranged from 2.42 to 2.92. The Ws statistic for comparison with North-American databases significantly improved after exclusion of patients with these injuries. The Ws statistic for comparison with a British reference database remained unchanged. Hip fractures, minor pelvic fractures, femur fractures, and minor thoracic wall injuries are associated with increased mortality. Comparative outcome analysis of a population with a reference database that differs in case-mix with respect to these injuries should be interpreted cautiously. Prognostic study, level II.

Comparison of patient-reported outcome measures following total and unicondylar knee replacement

Following arthroplasty of the knee, the patient's perception of improvement in symptoms is fundamental to the assessment of outcome. Better clinical outcome may offset the inferior survival observed for some types of implant. By examining linked National Joint Registry (NJR) and patient-reported outcome measures (PROMs) data, we aimed to compare PROMs collected at a minimum of six months post-operatively for total (TKR: n = 23,393) and unicondylar knee replacements (UKR: n = 505). Improvements in knee-specific (Oxford knee score, OKS) and generic (EuroQol, EQ-5D) scores were compared and adjusted for case-mix differences using multiple regression. Whereas the improvements in the OKS and EQ-5D were significantly greater for TKR than for UKR, once adjustments were made for case-mix differences and pre-operative score, the improvements in the two scores were not significantly different. The adjusted mean differences in the improvement of OKS and EQ-5D were 0.0 (95% confidence interval (CI) -0.9 to 0.9; p = 0.96) and 0.009 (95% CI -0.034 to 0.015; p = 0.37), respectively. We found no difference in the improvement of either knee-specific or general health outcomes between TKR and UKR in a large cohort of registry patients. With concerns about significantly higher revision rates for UKR observed in worldwide registries, we question the widespread use of an arthroplasty that does not confer a significant benefit in clinical outcome.

The European COPD Audit: brothers in arms

In March 2010, the European Respiratory Society funded a pilot audit programme of the process and outcomes of hospital care for patients admitted with exacerbations of COPD and on the organisation of care at hospital level. Originally it was intended to involve just five countries in this pilot scheme but such was the enthusiasm from national societies that 13 participated in the data collection programme. In March 2012, ERS published a document containing the results at national level for 19,021 cases entered into the audit database and subsequently site specific data has been distributed to each of the 422 hospitals that participated as shown in figure 1. Figure 1 Hospitals participating in the European COPD Audit. These data are unique in that they have been collected by clinicians prospectively identifying cases then recording in hospital care processes and patient outcomes at 90 days after admission date. In order to collect the data, many individuals had to use their time and skills but the success of the data collection is testimony to the enthusiasm of clinicians to better understand the quality of care delivered in their own institutions. The results are salutary and deserve our attention. In essence, they describe a standard of care with significant variation in adherence to guideline standards both within countries and between them. Arterial blood gases are taken on admission in only 12% cases in one country ranging to 96% in another whilst antibiotics are given to only 54% cases in country A yet 95% in country B. Unsurprisingly, outcomes vary equally widely with median length of stay at national level varying 5–15 days. At this stage in the data analysis, it is not possible to say why this occurs and there may be a number of explanations. There are certainly significant differences in case mix between …

A Two‐Compartment Mixed‐Effects Gamma Regression Model for Quantifying Between‐Unit Variability in Length of Stay among Children Admitted to Intensive Care

To quantify between-unit variability in mean length of stay (LoS) between intensive care units (ICUs) after adjusting for differences in case mix using a method that does not require arbitrary trimming of data. An analysis of registry data from pediatric ICUs (PICUs) in Australia and New Zealand. The relationships between patient LoS and associated patient factors were modeled as a log-linear function of the covariates using two gamma distributions. The predicted distribution is estimated as a weighted average of the two distributions where the relative weighting is conditional on the patient's elective status. Data for 12,763 admissions submitted to the Australian and New Zealand Paediatric Intensive Care Registry from the eight dedicated PICUs in Australia and New Zealand in 2007 and 2008. The two distributions of the mixture model accurately described the distribution of short- and long-stay patients in ICUs. After adjusting for patient case mix, several sites had a statistically significant effect on patient LoS. The two-compartment model characterizes ICU LoS for short- and long-stay patients more effectively than a single-compartment model. There is significant site-level variation in the LoS among children admitted to ICUs in Australia and New Zealand. Differences in the site-level variation between short- and long-stay patients indicate differences in discharge practice.

Clinical course and outcome for critically ill children with Down syndrome: a retrospective cohort study

Children with Down syndrome (DS) have several genetic anomalies within chromosome 21 which may influence their response to critical illness. We compared the intensive care course and outcome of children with DS versus those without. Retrospective cohort study in four English paediatric intensive care units (ICUs) (2003-2009, n = 33,485). We examined, via a competing risks model, whether risk (subhazard) for ICU mortality differed for children with DS, after adjusting for important confounders. DS patients exhibited lower disease severity at ICU admission but subsequently required a higher proportion of cardiovascular support, and similar renal support to non-DS patients. Children with DS (n = 1,278) had lower crude mortality than those without (4.2 versus 6.2 %, p = 0.003). This was not significant when expressed as standardized mortality ratio: 0.83 [95 % confidence interval (CI) 0.63-1.09] versus 0.90 (95 % CI 0.86-0.94). However, the competing risks model showed that mortality risk was influenced by length of ICU stay. At admission, DS patients exhibited a subhazard for mortality of 0.63 (95 % CI 0.46-0.85), which increased to 1.00 by day 10 of admission, and continued rising above that of non-DS children thereafter. Children with DS require a higher proportion of organ support than expected by disease severity at ICU admission. In addition, the mortality risk for children with DS is dependent upon length of ICU stay. These findings could reflect differences in case mix, but are also compatible with different response to critical illness in this group.

Influence of hospital type on outcomes after oesophageal and gastric cancer surgery

Outcomes after oesophagectomy and gastrectomy vary considerably between hospitals. Possible explanations include differences in case mix, hospital volume and hospital type. The present study examined the distribution of oesophagectomies and gastrectomies between hospital types in the Netherlands, and the relationship between hospital type and outcome. Data were obtained from the nationwide Netherlands Cancer Registry. Hospitals were categorized as university hospitals (UH), non-university teaching hospitals (NUTH) and non-university non-teaching hospitals (NUNTH). Hospital type-outcome relationships were analysed by Cox regression, adjusting for case mix, hospital volume, year of diagnosis and use of multimodal therapies. Between 1989 and 2009, 10 025 oesophagectomies and 14 221 gastrectomies for cancer were performed in the Netherlands. The percentage of oesophagectomies and gastrectomies performed in UH increased from 17·6 and 6·4 per cent respectively in 1989 to 44·1 and 12·9 per cent in 2009. After oesophagectomy, the 3-month mortality rate was 2·5 per cent in UH, 4·4 per cent in NUTH and 4·1 per cent in NUNTH (P = 0·006 for UH versus NUTH). After gastrectomy, the 3-month mortality rate was 4·9 per cent in UH, 8·9 per cent in NUTH and 8·7 per cent in NUNTH (P < 0·001 for UH versus NUTH). Three-year survival was also higher in UH than in NUTH and NUNTH. Oesophagogastric resections performed in UH were associated with better outcomes but, owing to variation in outcomes within hospital types, centres of excellence cannot be designated solely on hospital type. Detailed information on case mix and outcomes is needed to identify centres of excellence.

Prediction of outcome after moderate and severe traumatic brain injury

The International Mission on Prognosis and Analysis of Clinical Trials and Corticoid Randomisation After Significant Head injury prognostic models predict outcome after traumatic brain injury but have not been compared in large datasets. The objective of this is study is to validate externally and compare the International Mission on Prognosis and Analysis of Clinical Trials and Corticoid Randomisation after Significant Head injury prognostic models for prediction of outcome after moderate or severe traumatic brain injury. External validation study. We considered five new datasets with a total of 9,036 patients, comprising three randomized trials and two observational series, containing prospectively collected individual traumatic brain injury patient data. Outcomes were mortality and unfavorable outcome, based on the Glasgow Outcome Score at 6 months after injury. To assess performance, we studied the discrimination of the models (by area under the receiver operating characteristic curves), and calibration (by comparison of the mean observed to predicted outcomes and calibration slopes). The highest discrimination was found in the Trauma Audit and Research Network trauma registry (area under the receiver operating characteristic curves between 0.83 and 0.87), and the lowest discrimination in the Pharmos trial (area under the receiver operating characteristic curves between 0.65 and 0.71). Although differences in predictor effects between development and validation populations were found (calibration slopes varying between 0.58 and 1.53), the differences in discrimination were largely explained by differences in case mix in the validation studies. Calibration was good, the fraction of observed outcomes generally agreed well with the mean predicted outcome. No meaningful differences were noted in performance between the International Mission on Prognosis and Analysis of Clinical Trials and Corticoid Randomisation After Significant Head injury models. More complex models discriminated slightly better than simpler variants. Since both the International Mission on Prognosis and Analysis of Clinical Trials and the Corticoid Randomisation After Significant Head injury prognostic models show good generalizability to more recent data, they are valid instruments to quantify prognosis in traumatic brain injury.